Microchip technology with electron transparent membranes is a key component for in situ liquid transmission electron microscope (TEM) characterization. The membranes can significantly influence the TEM imaging spatial resolution, not only due to introducing additional material layers but also due to the associated bulging. The membrane bulging is largely defined by the membrane materials, thickness, and short dimension. The impact of the membrane on the spatial resolution, especially the extent of its bulging, was systematically investigated through the impact on the signal-to-noise ratio, chromatic aberration, and beam broadening. The optimization of the membrane parameters is the key component when designing the in situ TEM liquid cell. The optimal membrane thickness of 50 nm was found which balances the impact of membrane bulging and membrane thickness. Beyond this, the short membrane window dimension and the chip nominal spacing should be minimized. However, these two parameters have practical limitations in regards to chip handling.

The spatial resolution of aberration-corrected annular dark field scanning transmission electron microscopy was studied as function of the vertical position z within a sample. The samples consisted of gold nanoparticles (AuNPs) positioned in different horizontal layers within aluminum matrices of 0.6 and 1.0 µm thickness. The highest resolution was achieved in the top layer, whereas the resolution was reduced by beam broadening for AuNPs deeper in the sample. To examine the influence of the beam broadening, the intensity profiles of line scans over nanoparticles at a certain vertical location were analyzed. The experimental data were compared with Monte Carlo simulations that accurately matched the data. The spatial resolution was also calculated using three different theoretical models of the beam blurring as function of the vertical position within the sample. One model considered beam blurring to occur as a single scattering event but was found to be inaccurate for larger depths of the AuNPs in the sample. Two models were adapted and evaluated that include estimates for multiple scattering, and these described the data with sufficient accuracy to be able to predict the resolution. The beam broadening depended on z1.5 in all three models.

The lateral and axial resolution of three-dimensional (3D) focal series aberration-corrected scanning transmission electron microscopy was studied for samples of different thicknesses. The samples consisted of gold nanoparticles placed on the top and at the bottom of silicon nitride membranes of thickness between 50 and 500 nm. Atomic resolution was obtained for nanoparticles on top of 50-, 100-, and 200-nm-thick membranes with respect to the electron beam traveling downward. Atomic resolution was also achieved for nanoparticles placed below 50-, 100-, and 200-nm-thick membranes but with a lower contrast at the larger thicknesses. Beam broadening led to a reduced resolution for a 500-nm-thick membrane. The influence of the beam broadening on the axial resolution was also studied using Monte Carlo simulations with a 3D sample geometry.

A focused ion beam (FIB) interface attached to a column of 200 keV transmission electron microscope (TEM) was developed for in situ micropatterning to semiconductors. TEM specimens of Si and GaAs, and those of a thin Ni2Si layer on a Si substrate were micromilled in the TEM during observation. A set of 6 x 6-um squares and alphabet letters were patterned with a 25 keV Ga+-FIB of 0.2-μm beam diameter at room temperature. The effect of FIB irradiation on the structural evolution was observed simultaneously by a TV-rate video camera and sequentially by regular film. FIB micropatterning to semiconductor specimens caused amorphization and Ga injection. The excess Ga in the specimens precipitated as metastable solid γ-phase for Si and as liquid phase for GaAs. Ni2Si/Si specimens lost silicide crystallinity after FIB patterning. Annealing of these bilayer specimens at 673K resulted in the precipitation of Ni-rich silicide.